There is very little information about ISSE available, but as a starting point, watch https://www.youtube.com/watch?v=tKAX5pNPQp8&list=RDRd3prIkO5bg&index=3. The big difference with ISSE (and other tools that base on AI and/or ML, like some experiments with TensorFlow) in comparison to our standard tools like RX, WL or SLP is the fact that they don’t need surgical instructions from the user. When defining what you want to extract, just roughly mark the corresponding areas in both the time and the pitch domain in the spectrogram and see what happens.
One might say “hell, how can this be useful if I cannot clearly predict what will happen to my audio?!”. Good point, but I don’t care if it just works. Included with the ISSE download comes a speech file with a ringing cell phone. The ringtone covers two different frequencies that play quickly after each other, and they both show some harmonics. Try to remove this with RX, you will probably have no success. With ISSE, you end up with two files (speech and cell phone) that sound as if they were never mixed together 